Digital Predistortion Technology Based On Compressed Sensing

With the popularization of 5G communication system,the transmission rate of communication signals becomes higher and higher.High-speed transmission signals are often accompanied by the characteristics of high peak-to-average ratio and large bandwidth,resulting in more complex nonlinear characteristics of power amplifiers in wireless communication systems.The nonlinear distortion of the power amplifier has become one of the bottlenecks restricting the performance of the wireless communication system.With the advantages of superior performance,strong adaptability,stable work,and low implementation cost,digital predistortion technology has become the most popular power amplifier linearization technology.The various models currently used for digital predistortion have problems such as a large number of coefficients,high computational complexity,and numerical instability.On the other hand,the existing analog-to-digital converters can not meet the requirements of high speed sampling of large bandwidth signals in predistortion systems.In response to the abovementioned needs,this article uses compressed sensing theory to study the optimization method of digital predistortion technology.In the dual-band scenario,the number of pre-distortion model coefficients is often very large and the calculation is complicated.Therefore,the pre-distortion model needs to be simplified and reduced in dimension.In this paper,the orthogonal matching pursuit algorithm in the compressed sensing field is used to reduce the dimensionality of the predistortion model matrix.Considering that the premise of its application is the known signal sparsity,a model selection strategy is designed according to the Bayesian information criterion.The dimensionality reduction can be achieved when the sparsity is unknown.In order to more accurately describe the nonlinear characteristics of signals in different amplitude intervals,the vector quantization method is combined with the model dimensionality reduction method,and a piecewise dimensionality reduction adaptive digital predistortion method based on orthogonal matching pursuit is proposed.Simulation and actual measurement results show that this method reduces the number of predistortion model coefficients to 48%of the traditional predistortion method while maintaining accuracy.When the sampling rate of the feedback loop of the predistortion system is too high,the analog-to-digital converter that can meet the requirements is expensive and consumes too much power.In order to overcome this problem,based on the sparseness of the signal,the random demodulation structure is combined with the compressed sensing algorithm,the signal is multiplied by the pseudo-random sequence,and the useful information in the signal is preserved.After sampling at a rate lower than Nyquist rate,the signal is recovered and reconstructed by orthogonal matching pursuit algorithm.The method is simulated and verified,and it can still maintain a good recovery effect and linearization effect when the sampling rate is 20%of the Nyquist sampling rate.